This markdown represents a summary of the functional connectivity (FC) results, i.e. how changes in resting state functional connectivity (RSFC) relate to memory performance. This captures post-encoding/consolition processes. Addionally, the concatenated time course of the task data was used to investigate functional connectivity during online and offline encoding and how this relates to memory performance.

Behavioural scores

Measures of Learning

Total # items encoded

To calculate the total # of items encoded, for each subject, a sum score was created adding up all items remembered

  1. regardless of threshold to test whether cognition performance was above chance level
# one sample t-test to see whether recognition performance was above chance
describe(dfWide$allConf_abs)
##    vars  n  mean   sd median trimmed  mad min max range skew kurtosis   se
## X1    1 50 22.54 4.12     22   22.43 4.45  15  32    17  0.2     -0.6 0.58
t.test(dfWide$allConf_abs, mu = (length(unique(dfLong$stimID))/4), alternative = "greater")
## 
##  One Sample t-test
## 
## data:  dfWide$allConf_abs
## t = 23.257, df = 49, p-value < 2.2e-16
## alternative hypothesis: true mean is greater than 9
## 95 percent confidence interval:
##  21.56393      Inf
## sample estimates:
## mean of x 
##     22.54
effsize::cohen.d(dfWide$allConf_abs, f = NA, mu = (length(unique(dfLong$stimID))/4))
## 
## Cohen's d (single sample)
## 
## d estimate: 3.28903 (large)
## Reference mu: 9
## 95 percent confidence interval:
##    lower    upper 
## 2.417288 4.160771
  1. at the high confidence threshold to calculate descriptive statistics and test for effects of group.
# two sample t-test to see whether high confidence recognition differed between both groups
describe(dfWide$highConf_abs)
##    vars  n  mean   sd median trimmed  mad min max range  skew kurtosis   se
## X1    1 50 15.14 5.35     15   15.15 4.45   3  27    24 -0.07    -0.13 0.76
describe.by(dfWide$highConf_abs, dfWide$group)
## Warning: describe.by is deprecated. Please use the describeBy function
## 
##  Descriptive statistics by group 
## group: cont
##    vars  n  mean   sd median trimmed  mad min max range  skew kurtosis   se
## X1    1 25 15.52 4.65     15   15.57 4.45   3  25    22 -0.23      0.6 0.93
## ------------------------------------------------------------ 
## group: exp
##    vars  n  mean   sd median trimmed  mad min max range skew kurtosis   se
## X1    1 25 14.76 6.04     16   14.57 5.93   4  27    23 0.08    -0.72 1.21
t.test(dfWide$highConf_abs ~ dfWide$group)
## 
##  Welch Two Sample t-test
## 
## data:  dfWide$highConf_abs by dfWide$group
## t = 0.49846, df = 45.028, p-value = 0.6206
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -2.310821  3.830821
## sample estimates:
## mean in group cont  mean in group exp 
##              15.52              14.76
effsize::cohen.d(dfWide$allConf_abs, f = dfWide$group)
## 
## Cohen's d
## 
## d estimate: 0.4230503 (small)
## 95 percent confidence interval:
##      lower      upper 
## -0.1519691  0.9980698

Corrected memory score

To follow suggestions by others investigating post-encoding rest (e.g., Duncan et al., 2014, Tompary et al., 2015), a corrected high confidence memory score was calculated to account for correct guesses using the formula below:
Corrected memory score = sum(high condidence correct) - sum ((high confidence incorrect) / number of alternatives)

# two sample t-test to see whether corrected high confidence recognition differed between both groups
describe(dfWide$memory_corrected)
##    vars  n  mean   sd median trimmed  mad  min   max range skew kurtosis   se
## X1    1 50 13.86 5.42  13.88   13.81 4.45 2.75 26.25  23.5 0.05    -0.33 0.77
describe.by(dfWide$memory_corrected, dfWide$group)
## Warning: describe.by is deprecated. Please use the describeBy function
## 
##  Descriptive statistics by group 
## group: cont
##    vars  n  mean   sd median trimmed  mad  min max range skew kurtosis   se
## X1    1 25 14.26 4.72     14   14.14 4.45 2.75  24 21.25 0.05      0.1 0.94
## ------------------------------------------------------------ 
## group: exp
##    vars  n  mean   sd median trimmed  mad  min   max range skew kurtosis   se
## X1    1 25 13.45 6.11  13.75    13.2 7.41 3.75 26.25  22.5 0.13    -0.82 1.22
t.test(dfWide$memory_corrected ~ dfWide$group)
## 
##  Welch Two Sample t-test
## 
## data:  dfWide$memory_corrected by dfWide$group
## t = 0.52492, df = 45.122, p-value = 0.6022
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -2.297721  3.917721
## sample estimates:
## mean in group cont  mean in group exp 
##              14.26              13.45
effsize::cohen.d(dfWide$memory_corrected, f = dfWide$group)
## 
## Cohen's d
## 
## d estimate: 0.1484696 (negligible)
## 95 percent confidence interval:
##      lower      upper 
## -0.4210068  0.7179459

Measures of the Effects of Curiosity on Learning

Within-person correlation between curiosity and encoding

For each subject, the correlation between mean-centred curiosity ratings and high confidence recogntion is computed and correlation coefficients are Fisher’s z-transformed.

# two sample t-test to see whether within-person correlation (z transformed) differed between both groups
describe(dfWide$cor_cur_mem_z)
##    vars  n mean   sd median trimmed  mad   min  max range  skew kurtosis   se
## X1    1 50 0.03 0.19   0.05    0.05 0.19 -0.45 0.34  0.79 -0.56     -0.1 0.03
describe.by(dfWide$cor_cur_mem_z, dfWide$group)
## Warning: describe.by is deprecated. Please use the describeBy function
## 
##  Descriptive statistics by group 
## group: cont
##    vars  n mean   sd median trimmed mad   min  max range  skew kurtosis   se
## X1    1 25 0.06 0.21   0.06    0.08 0.2 -0.45 0.34  0.79 -0.81    -0.04 0.04
## ------------------------------------------------------------ 
## group: exp
##    vars  n mean   sd median trimmed  mad  min  max range  skew kurtosis   se
## X1    1 25 0.01 0.17  -0.01    0.02 0.17 -0.4 0.33  0.73 -0.25     -0.3 0.03
t.test(dfWide$cor_cur_mem_z ~ dfWide$group)
## 
##  Welch Two Sample t-test
## 
## data:  dfWide$cor_cur_mem_z by dfWide$group
## t = 0.84204, df = 46.073, p-value = 0.4041
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.06250556  0.15241609
## sample estimates:
## mean in group cont  mean in group exp 
##         0.05632509         0.01136982
effsize::cohen.d(dfWide$cor_cur_mem_z, f = dfWide$group)
## 
## Cohen's d
## 
## d estimate: 0.2381651 (small)
## 95 percent confidence interval:
##      lower      upper 
## -0.3325409  0.8088710

Curiosity-driven memory enhancement (CDMB)

Previous research (Gruber et al., 2014) has quantified the effects of curiosity on memory using an index referred to as curiosity-motivated learning index, comparing the number of items later remembered vs. forgotten in states of high vs. low curiosity.

\[ CDMB = {\sum_{i=1}^{36} curiosity_{CWC} > 0 * memory_{Dummy} \over \sum_{i=1}^{36} curiosity_{CWC} > 0 \ } - {\sum_{i=1}^{36} curiosity_{CWC} < 0 * memory_{Dummy} \over \sum_{i=1}^{36} curiosity_{CWC} < 0 \ } \]

As such, the CDMB computes the ratio between items remembered eliciting high ratings of curiosity to the total number of items eliciting high curiosity ratings and campares it to the ratio of items remembered eliciting low ratings of curiosity compared to the total number of items receiving low curiosity ratings.

To determine high vs low ratings of curiosity, curiosity was centred within cluster (CWC) and compared against 0.

When computing the product between curiosity and memory encoding (dummy-coded), mean-centred curiosity ratings can be used as a continuous variable or as a binary variable. In either way, the resulting CDMB are highly correlated with each other (r < .90). To ease the understanding, a binary formulation of mean-centred curiosity is used, allowing for the following definition of CDMB:

“To compute the effect of curiosity on encoding, the curiosity-driven memory benefit index was computed, i.e., the difference in high confidence recognition memory performance between magic tricks eliciting high ratings of curiosity relative to the total number of magic tricks eliciting high ratings of curiosity and the high confidence recognition memory performance between magic tricks eliciting low ratings of curiosity relative to the total number of magic tricks eliciting low ratings of curiosity.”

# two sample t-test to see whether CDMB (dichotomous) differed between both groups
describe(dfWide$CDMB)
##    vars  n mean   sd median trimmed  mad   min  max range  skew kurtosis   se
## X1    1 50 0.02 0.22   0.04    0.04 0.17 -0.86 0.36  1.22 -1.31      3.2 0.03
describe.by(dfWide$CDMB, dfWide$group)
## Warning: describe.by is deprecated. Please use the describeBy function
## 
##  Descriptive statistics by group 
## group: cont
##    vars  n mean   sd median trimmed  mad   min  max range  skew kurtosis   se
## X1    1 25 0.08 0.21    0.1     0.1 0.22 -0.43 0.36  0.79 -0.74    -0.26 0.04
## ------------------------------------------------------------ 
## group: exp
##    vars  n  mean   sd median trimmed  mad   min max range  skew kurtosis   se
## X1    1 25 -0.05 0.22  -0.01   -0.02 0.17 -0.86 0.2  1.06 -2.02      5.2 0.04
t.test(dfWide$CDMB ~ dfWide$group)
## 
##  Welch Two Sample t-test
## 
## data:  dfWide$CDMB by dfWide$group
## t = 2.139, df = 47.908, p-value = 0.03756
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.007703609 0.249235574
## sample estimates:
## mean in group cont  mean in group exp 
##         0.08333497        -0.04513463
effsize::cohen.d(dfWide$CDMB, f = dfWide$group)
## 
## Cohen's d
## 
## d estimate: 0.6050002 (medium)
## 95 percent confidence interval:
##     lower     upper 
## 0.0234426 1.1865579
# re-run t test after rmoving outlier
q25 <- as.numeric(quantile(dfWide$CDMB)[2])
q75 <- as.numeric(quantile(dfWide$CDMB)[4])
iqr <-  q75 - q25# calculate IQR
tmp <- subset(dfWide, CDMB >= q25 - 1.5*iqr & CDMB <= q75 + 1.5*iqr) # this removes sub-control049 and sub-experimental018
t.test(tmp$CDMB ~ tmp$group)
## 
##  Welch Two Sample t-test
## 
## data:  tmp$CDMB by tmp$group
## t = 2.4882, df = 43.029, p-value = 0.01679
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.02199897 0.21017062
## sample estimates:
## mean in group cont  mean in group exp 
##         0.10478384        -0.01130095
effsize::cohen.d(tmp$CDMB, f = tmp$group)
## 
## Cohen's d
## 
## d estimate: 0.7182767 (medium)
## 95 percent confidence interval:
##     lower     upper 
## 0.1187597 1.3177936

Relationship between the measures

Below, all behavioural indices described above are combined in a scatter plot matrix.

While measures quantifying the same construct are highly correlated, measures of different constructs are uncorrelated.

Pearson's product-moment correlation

data: dfWide\(highConf_abs and dfWide\)memory_corrected t = 40.492, df = 48, p-value < 2.2e-16 alternative hypothesis: true correlation is not equal to 0 95 percent confidence interval: 0.9747652 0.9918884 sample estimates: cor 0.9856758

Pearson's product-moment correlation

data: dfWide\(CDMB and dfWide\)cor_cur_mem_z t = 11.3, df = 48, p-value = 3.987e-15 alternative hypothesis: true correlation is not equal to 0 95 percent confidence interval: 0.7528223 0.9139862 sample estimates: cor 0.8525262

Functional connectivity during each phase

t-Tests for FC values

# pre learning
mean(dfWide$FC_pre_s4)
## [1] 0.0347034
sd(dfWide$FC_pre_s4)
## [1] 0.02516238
# post learning
mean(dfWide$FC_post_s4)
## [1] 0.03878373
sd(dfWide$FC_post_s4)
## [1] 0.03042794
# diff
mean(dfWide$FC_diff_s4)
## [1] 0.004080335
sd(dfWide$FC_diff_s4)
## [1] 0.030998
# RSFC change across whole sample
t.test(dfWide[, "FC_diff_s4"], alternative = "greater")
## 
##  One Sample t-test
## 
## data:  dfWide[, "FC_diff_s4"]
## t = 0.93078, df = 49, p-value = 0.1783
## alternative hypothesis: true mean is greater than 0
## 95 percent confidence interval:
##  -0.003269295          Inf
## sample estimates:
##   mean of x 
## 0.004080335
effsize::cohen.d(dfWide$FC_diff_s4, f = NA)
## 
## Cohen's d (single sample)
## 
## d estimate: 0.1316322 (negligible)
## Reference mu: 0
## 95 percent confidence interval:
##      lower      upper 
## -0.4373767  0.7006411
# RSFC between both groups
t.test(dfWide[, "FC_diff_s4"] ~ dfWide$group)
## 
##  Welch Two Sample t-test
## 
## data:  dfWide[, "FC_diff_s4"] by dfWide$group
## t = 0.78748, df = 47.816, p-value = 0.4349
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.01076769  0.02463001
## sample estimates:
## mean in group cont  mean in group exp 
##       0.0075459160       0.0006147536
sd(dfWide$FC_diff_s4[dfWide$group == "cont"])
## [1] 0.03206868
sd(dfWide$FC_diff_s4[dfWide$group == "exp"])
## [1] 0.03013914
effsize::cohen.d(dfWide$FC_diff_s4, f = dfWide$group)
## 
## Cohen's d
## 
## d estimate: 0.2227318 (small)
## 95 percent confidence interval:
##      lower      upper 
## -0.3477221  0.7931858

Brain-behaviour correlations

total # items encoded

cor <- "pearson"
# whole sample
if (cor == "spearman") {
  DescTools::SpearmanRho(dfWide[, "highConf_abs"], dfWide[, "FC_diff_s4"], conf.level = 0.95)
} else {
  cor.test(dfWide[, "highConf_abs"], dfWide[, "FC_diff_s4"])
}
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[, "highConf_abs"] and dfWide[, "FC_diff_s4"]
## t = 0.55646, df = 48, p-value = 0.5805
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2028063  0.3505956
## sample estimates:
##        cor 
## 0.08006087
# control
if (cor == "spearman") {
  cor_cont <- DescTools::SpearmanRho(dfWide[dfWide$group == "cont", "highConf_abs"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_cont <- cor.test(dfWide[dfWide$group == "cont", "highConf_abs"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])
}
cor_cont
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "cont", "highConf_abs"] and dfWide[dfWide$group == "cont", paste0("FC_diff_s4")]
## t = -1.5778, df = 23, p-value = 0.1283
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.6298669  0.0942553
## sample estimates:
##        cor 
## -0.3125146
# experimental
if (cor == "spearman") {
  cor_exp <- DescTools::SpearmanRho(dfWide[dfWide$group == "exp", "highConf_abs"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_exp <- cor.test(dfWide[dfWide$group == "exp", "highConf_abs"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])
}
cor_exp
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "exp", "highConf_abs"] and dfWide[dfWide$group == "exp", paste0("FC_diff_s4")]
## t = 2.0276, df = 23, p-value = 0.05434
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.006754121  0.679926444
## sample estimates:
##      cor 
## 0.389416
# group effect
if (cor == "spearman") {
  cocor::cocor.indep.groups(cor_exp[1], cor_cont[1], 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
} else {
  cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
}
## 
##   Results of a comparison of two correlations based on independent groups
## 
## Comparison between r1.jk = 0.3894 and r2.hm = -0.3125
## Difference: r1.jk - r2.hm = 0.7019
## Group sizes: n1 = 25, n2 = 25
## Null hypothesis: r1.jk is equal to r2.hm
## Alternative hypothesis: r1.jk is not equal to r2.hm (two-sided)
## Alpha: 0.05
## 
## fisher1925: Fisher's z (1925)
##   z = 2.4359, p-value = 0.0149
##   Null hypothesis rejected
## 
## zou2007: Zou's (2007) confidence interval
##   95% confidence interval for r1.jk - r2.hm: 0.1341 1.1322
##   Null hypothesis rejected (Interval does not include 0)

Curiosity-driven memory benefit

cor <- "pearson"
# whole sample
if (cor == "spearman") {
  DescTools::SpearmanRho(dfWide[, "CDMB"], dfWide[, "FC_diff_s4"], conf.level = 0.95)
} else {
  cor.test(dfWide[, "CDMB"], dfWide[, "FC_diff_s4"])
}
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[, "CDMB"] and dfWide[, "FC_diff_s4"]
## t = 0.51151, df = 48, p-value = 0.6113
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2090009  0.3449091
## sample estimates:
##        cor 
## 0.07363021
# control
if (cor == "spearman") {
  cor_cont <- DescTools::SpearmanRho(dfWide[dfWide$group == "cont", "CDMB"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_cont <- cor.test(dfWide[dfWide$group == "cont", "CDMB"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])
}
cor_cont
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "cont", "CDMB"] and dfWide[dfWide$group == "cont", paste0("FC_diff_s4")]
## t = -0.87836, df = 23, p-value = 0.3888
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.5370550  0.2314527
## sample estimates:
##       cor 
## -0.180154
# experimental
if (cor == "spearman") {
  cor_exp <- DescTools::SpearmanRho(dfWide[dfWide$group == "exp", "CDMB"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_exp <- cor.test(dfWide[dfWide$group == "exp", "CDMB"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])
}
cor_exp
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "exp", "CDMB"] and dfWide[dfWide$group == "exp", paste0("FC_diff_s4")]
## t = 1.3409, df = 23, p-value = 0.193
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.1408479  0.6005078
## sample estimates:
##       cor 
## 0.2692687
# group effect
if (cor == "spearman") {
  cocor::cocor.indep.groups(cor_exp[1], cor_cont[1], 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
} else {
  cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
  r_cmle <- cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
}

Figure 3

corrected memory

# whole sample
cor.test(dfWide[, "memory_corrected"], dfWide[, "FC_diff_s4"])
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[, "memory_corrected"] and dfWide[, "FC_diff_s4"]
## t = 0.45423, df = 48, p-value = 0.6517
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2168751  0.3376215
## sample estimates:
##        cor 
## 0.06542191
# control
cor_cont <- cor.test(dfWide[dfWide$group == "cont", "memory_corrected"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])

# experimental
cor_exp <- cor.test(dfWide[dfWide$group == "exp", "memory_corrected"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])

# group effect
cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                          test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                          data.name = NULL, var.labels = NULL, return.htest = FALSE)
## 
##   Results of a comparison of two correlations based on independent groups
## 
## Comparison between r1.jk = 0.3199 and r2.hm = -0.2603
## Difference: r1.jk - r2.hm = 0.5802
## Group sizes: n1 = 25, n2 = 25
## Null hypothesis: r1.jk is equal to r2.hm
## Alternative hypothesis: r1.jk is not equal to r2.hm (two-sided)
## Alpha: 0.05
## 
## fisher1925: Fisher's z (1925)
##   z = 1.9834, p-value = 0.0473
##   Null hypothesis rejected
## 
## zou2007: Zou's (2007) confidence interval
##   95% confidence interval for r1.jk - r2.hm: 0.0028 1.0393
##   Null hypothesis rejected (Interval does not include 0)

within-person correlation

# whole sample
cor.test(dfWide[, "cor_cur_mem_z"], dfWide[, "FC_diff_s4"])
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[, "cor_cur_mem_z"] and dfWide[, "FC_diff_s4"]
## t = -0.64375, df = 48, p-value = 0.5228
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.3615560  0.1907405
## sample estimates:
##         cor 
## -0.09251931
# control
cor_cont <- cor.test(dfWide[dfWide$group == "cont", "cor_cur_mem_z"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])
cor_cont
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "cont", "cor_cur_mem_z"] and dfWide[dfWide$group == "cont", paste0("FC_diff_s4")]
## t = -1.4352, df = 23, p-value = 0.1647
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.6124445  0.1222925
## sample estimates:
##        cor 
## -0.2866918
# experimental
cor_exp <- cor.test(dfWide[dfWide$group == "exp", "cor_cur_mem_z"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])
cor_exp
## 
##  Pearson's product-moment correlation
## 
## data:  dfWide[dfWide$group == "exp", "cor_cur_mem_z"] and dfWide[dfWide$group == "exp", paste0("FC_diff_s4")]
## t = 0.61013, df = 23, p-value = 0.5478
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  -0.2830414  0.4965722
## sample estimates:
##       cor 
## 0.1262039
# group effect
cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                          test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                          data.name = NULL, var.labels = NULL, return.htest = FALSE)
## 
##   Results of a comparison of two correlations based on independent groups
## 
## Comparison between r1.jk = 0.1262 and r2.hm = -0.2867
## Difference: r1.jk - r2.hm = 0.4129
## Group sizes: n1 = 25, n2 = 25
## Null hypothesis: r1.jk is equal to r2.hm
## Alternative hypothesis: r1.jk is not equal to r2.hm (two-sided)
## Alpha: 0.05
## 
## fisher1925: Fisher's z (1925)
##   z = 1.3991, p-value = 0.1618
##   Null hypothesis retained
## 
## zou2007: Zou's (2007) confidence interval
##   95% confidence interval for r1.jk - r2.hm: -0.1657 0.9061
##   Null hypothesis retained (Interval includes 0)

Figure S1

total # items encoded - Spearman

cor <- "spearman"
# whole sample
if (cor == "spearman") {
  DescTools::SpearmanRho(dfWide[, "highConf_abs"], dfWide[, "FC_diff_s4"], conf.level = 0.95)
} else {
  cor.test(dfWide[, "highConf_abs"], dfWide[, "FC_diff_s4"])
}
## Registered S3 method overwritten by 'DescTools':
##   method        from       
##   print.palette wesanderson
##        rho     lwr.ci     upr.ci 
##  0.1495985 -0.1343433  0.4108387
# control
if (cor == "spearman") {
  cor_cont <- DescTools::SpearmanRho(dfWide[dfWide$group == "cont", "highConf_abs"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_cont <- cor.test(dfWide[dfWide$group == "cont", "highConf_abs"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])
}
cor_cont
##        rho     lwr.ci     upr.ci 
## -0.1753589 -0.5335222  0.2361337
# experimental
if (cor == "spearman") {
  cor_exp <- DescTools::SpearmanRho(dfWide[dfWide$group == "exp", "highConf_abs"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_exp <- cor.test(dfWide[dfWide$group == "exp", "highConf_abs"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])
}
cor_exp
##        rho     lwr.ci     upr.ci 
## 0.42606418 0.03719513 0.70286669
# group effect
if (cor == "spearman") {
  cocor::cocor.indep.groups(cor_exp[1], cor_cont[1], 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
} else {
  cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
}
## 
##   Results of a comparison of two correlations based on independent groups
## 
## Comparison between r1.jk = 0.4261 and r2.hm = -0.1754
## Difference: r1.jk - r2.hm = 0.6014
## Group sizes: n1 = 25, n2 = 25
## Null hypothesis: r1.jk is equal to r2.hm
## Alternative hypothesis: r1.jk is not equal to r2.hm (two-sided)
## Alpha: 0.05
## 
## fisher1925: Fisher's z (1925)
##   z = 2.0970, p-value = 0.0360
##   Null hypothesis rejected
## 
## zou2007: Zou's (2007) confidence interval
##   95% confidence interval for r1.jk - r2.hm: 0.0353 1.0541
##   Null hypothesis rejected (Interval does not include 0)

Curiosity-driven memory benefit - Spearman

cor <- "spearman"
# whole sample
if (cor == "spearman") {
  DescTools::SpearmanRho(dfWide[, "CDMB"], dfWide[, "FC_diff_s4"], conf.level = 0.95)
} else {
  cor.test(dfWide[, "CDMB"], dfWide[, "FC_diff_s4"])
}
##         rho      lwr.ci      upr.ci 
## -0.02458642 -0.30087508  0.25550990
# control
if (cor == "spearman") {
  cor_cont <- DescTools::SpearmanRho(dfWide[dfWide$group == "cont", "CDMB"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_cont <- cor.test(dfWide[dfWide$group == "cont", "CDMB"], dfWide[dfWide$group == "cont", paste0("FC_diff_s4")])
}
cor_cont
##        rho     lwr.ci     upr.ci 
## -0.1654164 -0.5261571  0.2457789
# experimental
if (cor == "spearman") {
  cor_exp <- DescTools::SpearmanRho(dfWide[dfWide$group == "exp", "CDMB"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")], conf.level = 0.95)
} else {
  cor_exp <- cor.test(dfWide[dfWide$group == "exp", "CDMB"], dfWide[dfWide$group == "exp", paste0("FC_diff_s4")])
}
cor_exp
##          rho       lwr.ci       upr.ci 
##  0.006923077 -0.389272686  0.400957149
# group effect
if (cor == "spearman") {
  cocor::cocor.indep.groups(cor_exp[1], cor_cont[1], 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
} else {
  cocor::cocor.indep.groups(cor_exp$estimate, cor_cont$estimate, 25, 25, alternative = "two.sided",
                            test = "all", alpha = 0.05, conf.level = 0.95, null.value = 0,
                            data.name = NULL, var.labels = NULL, return.htest = FALSE)
}
## 
##   Results of a comparison of two correlations based on independent groups
## 
## Comparison between r1.jk = 0.0069 and r2.hm = -0.1654
## Difference: r1.jk - r2.hm = 0.1723
## Group sizes: n1 = 25, n2 = 25
## Null hypothesis: r1.jk is equal to r2.hm
## Alternative hypothesis: r1.jk is not equal to r2.hm (two-sided)
## Alpha: 0.05
## 
## fisher1925: Fisher's z (1925)
##   z = 0.5767, p-value = 0.5642
##   Null hypothesis retained
## 
## zou2007: Zou's (2007) confidence interval
##   95% confidence interval for r1.jk - r2.hm: -0.3987 0.7066
##   Null hypothesis retained (Interval includes 0)

Brain-encoding correlations as a function of smoothing kernels

FC connectivity during consolidation (i.e. changes in RSFC between HPC & midbrain) and encoding (task FC between HPC & midbrain) were correlated with the behavioural measurements of learning. This was done separately for each smoothing kernel applied during pre-processing.

Below, a graph was created for each behavioural measurement of learning plotting the different memory levels on the y-axis and correlation coefficients and their 95% confidence interval on the x-axis (dashed line is equivalent to r = 0). Correlation values were plotted for the whole sample and seperately for control- and incentives group. Additionally, their difference was plotted. Different colours and shape indicate different smoothing kernels (see legend). Data from the consolidation was plotted on the left side, whereas data on the right stems from encoding. The data used in the plot can be found at the end of the paper in form from tables.

Figure S2

Correlation between FC and memory measurements for each smoothing kernel

Each table below shows the correlation between behavioural measures of learning and firstly, the change in RSFC in the context of consolidation and secondly, task-FC in the context of encoding.

note that all p < 0.05 were printed in bold whereas all 0.05 < p < 0.10 were printed in italics

Correlation between FC and behaviour at Change [pre < post]
Whole sample
Control group
Incentives group
Difference C < I
smooth cor pval cor_cont pval_cor_cont cor_exp pval_cor_exp corrDiff pval_corrDiff
Total # items encoded
FWHM = 0 0.069 0.633 -0.335 0.102 0.388 0.055 0.723 0.012
FWHM = 4 0.080 0.580 -0.313 0.128 0.389 0.054 0.702 0.015
FWHM = 6 0.134 0.353 -0.342 0.094 0.492 0.013 0.834 0.003
FWHM = 8 0.165 0.253 -0.328 0.110 0.536 0.006 0.864 0.002
Curiosity-driven memory benefit
FWHM = 0 0.058 0.690 -0.166 0.427 0.220 0.290 0.386 0.194
FWHM = 4 0.074 0.611 -0.180 0.389 0.269 0.193 0.449 0.129
FWHM = 6 0.113 0.435 -0.108 0.606 0.273 0.186 0.382 0.197
FWHM = 8 0.122 0.399 -0.092 0.662 0.279 0.177 0.371 0.209
Corrected memory
FWHM = 0 0.056 0.700 -0.277 0.180 0.316 0.124 0.594 0.042
FWHM = 4 0.065 0.652 -0.260 0.209 0.320 0.119 0.580 0.047
FWHM = 6 0.116 0.423 -0.302 0.142 0.428 0.033 0.730 0.011
FWHM = 8 0.143 0.321 -0.298 0.148 0.475 0.016 0.773 0.006
Within-person correlation
FWHM = 0 -0.092 0.524 -0.268 0.195 0.102 0.628 0.370 0.211
FWHM = 4 -0.093 0.523 -0.287 0.165 0.126 0.548 0.413 0.162
FWHM = 6 -0.061 0.673 -0.225 0.280 0.110 0.602 0.335 0.261
FWHM = 8 -0.049 0.738 -0.203 0.330 0.113 0.591 0.316 0.289
Correlation between FC and behaviour at Online encoding
Whole sample
Control group
Incentives group
Difference C < I
smooth cor pval cor_cont pval_cor_cont cor_exp pval_cor_exp corrDiff pval_corrDiff
Total # items encoded
FWHM = 0 -0.010 0.943 0.012 0.955 -0.053 0.801 -0.065 0.829
FWHM = 4 0.029 0.843 0.068 0.745 -0.028 0.895 -0.096 0.749
FWHM = 6 0.108 0.456 0.064 0.761 0.156 0.457 0.092 0.758
FWHM = 8 0.099 0.494 0.031 0.882 0.180 0.390 0.148 0.618
Curiosity-driven memory benefit
FWHM = 0 -0.117 0.417 -0.222 0.287 -0.092 0.660 0.129 0.660
FWHM = 4 -0.153 0.288 -0.288 0.162 -0.095 0.652 0.194 0.503
FWHM = 6 -0.131 0.364 -0.314 0.126 0.030 0.886 0.344 0.238
FWHM = 8 -0.094 0.517 -0.301 0.144 0.121 0.564 0.422 0.151
Corrected memory
FWHM = 0 -0.020 0.889 -0.004 0.987 -0.059 0.779 -0.056 0.853
FWHM = 4 0.023 0.875 0.060 0.776 -0.033 0.875 -0.093 0.757
FWHM = 6 0.105 0.469 0.063 0.765 0.149 0.476 0.086 0.772
FWHM = 8 0.097 0.504 0.034 0.874 0.171 0.414 0.137 0.645
Within-person correlation
FWHM = 0 -0.165 0.253 -0.164 0.433 -0.213 0.307 -0.049 0.867
FWHM = 4 -0.197 0.171 -0.225 0.279 -0.198 0.342 0.027 0.924
FWHM = 6 -0.194 0.177 -0.284 0.169 -0.073 0.729 0.211 0.468
FWHM = 8 -0.162 0.260 -0.292 0.156 0.044 0.833 0.337 0.252
Correlation between FC and behaviour at Offline encoding
Whole sample
Control group
Incentives group
Difference C < I
smooth cor pval cor_cont pval_cor_cont cor_exp pval_cor_exp corrDiff pval_corrDiff
Total # items encoded
FWHM = 0 0.041 0.778 -0.091 0.666 0.170 0.417 0.261 0.383
FWHM = 4 0.064 0.657 -0.047 0.822 0.166 0.428 0.213 0.476
FWHM = 6 0.106 0.463 0.005 0.980 0.196 0.349 0.190 0.522
FWHM = 8 0.102 0.483 0.009 0.965 0.185 0.376 0.176 0.555
Curiosity-driven memory benefit
FWHM = 0 0.009 0.950 -0.283 0.170 0.359 0.078 0.642 0.027
FWHM = 4 -0.015 0.916 -0.311 0.130 0.310 0.131 0.621 0.033
FWHM = 6 -0.086 0.553 -0.323 0.115 0.156 0.457 0.479 0.103
FWHM = 8 -0.112 0.437 -0.312 0.129 0.089 0.672 0.401 0.171
Corrected memory
FWHM = 0 0.062 0.671 -0.040 0.850 0.163 0.436 0.203 0.498
FWHM = 4 0.082 0.570 0.006 0.978 0.154 0.462 0.148 0.620
FWHM = 6 0.116 0.423 0.048 0.818 0.177 0.397 0.129 0.665
FWHM = 8 0.109 0.451 0.044 0.836 0.170 0.417 0.126 0.672
Within-person correlation
FWHM = 0 -0.086 0.552 -0.298 0.148 0.237 0.254 0.535 0.069
FWHM = 4 -0.093 0.520 -0.313 0.128 0.216 0.300 0.529 0.072
FWHM = 6 -0.126 0.385 -0.315 0.126 0.128 0.540 0.443 0.131
FWHM = 8 -0.150 0.298 -0.316 0.124 0.075 0.722 0.390 0.183

Predict behavioural measures of learning using FC values

Smoothing kernel FWHM = 0: Linear model predicting Total # items encoded

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 15.80 13.68 – 17.92 15.04 <0.001 15.52 13.57 – 17.47 16.58 <0.001 14.76 12.37 – 17.15 12.83 <0.001
Online FC -38.02 -127.09 – 51.05 -0.86 0.394 0.45 -108.70 – 109.60 0.01 0.993 -91.19 -254.82 – 72.44 -1.16 0.260
Offline FC 20.68 -44.56 – 85.91 0.64 0.526 -0.13 -83.98 – 83.72 -0.00 0.997 36.99 -73.31 – 147.28 0.70 0.493
Incentives -0.85 -3.85 – 2.16 -0.57 0.574
RSFC change -59.01 -133.41 – 15.40 -1.60 0.117 -52.49 -123.81 – 18.84 -1.53 0.141 95.19 0.73 – 189.64 2.10 0.048
Incentives * RSFC change 146.80 39.20 – 254.39 2.75 0.009
Observations 50 25 25
R2 / R2 adjusted 0.156 / 0.060 0.112 / -0.015 0.207 / 0.094
  Whole sample
Predictors Estimates CI Statistic p
Intercept 15.80 13.68 – 17.92 15.04 <0.001
Online FC -38.02 -127.09 – 51.05 -0.86 0.394
Offline FC 20.68 -44.56 – 85.91 0.64 0.526
Incentives -0.85 -3.85 – 2.16 -0.57 0.574
RSFC change -59.01 -133.41 – 15.40 -1.60 0.117
Incentives * RSFC change 146.80 39.20 – 254.39 2.75 0.009
Observations 50
R2 / R2 adjusted 0.156 / 0.060

Smoothing kernel FWHM = 0: Linear model predicting Curiosity-driven memory benefit

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.10 0.01 – 0.18 2.22 0.032 0.08 -0.00 – 0.17 1.98 0.061 -0.05 -0.13 – 0.04 -1.10 0.283
Online FC -2.89 -6.52 – 0.74 -1.61 0.115 -1.24 -6.15 – 3.67 -0.53 0.604 -3.91 -9.74 – 1.92 -1.40 0.177
Offline FC 1.02 -1.64 – 3.68 0.78 0.442 -1.29 -5.06 – 2.49 -0.71 0.486 3.67 -0.26 – 7.60 1.94 0.066
Incentives -0.14 -0.26 – -0.02 -2.31 0.026
RSFC change -1.51 -4.54 – 1.52 -1.00 0.322 -0.84 -4.05 – 2.36 -0.55 0.590 1.76 -1.60 – 5.13 1.09 0.288
Incentives * RSFC change 3.60 -0.78 – 7.99 1.66 0.105
Observations 50 25 25
R2 / R2 adjusted 0.171 / 0.077 0.101 / -0.028 0.220 / 0.109
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.10 0.01 – 0.18 2.22 0.032
Online FC -2.89 -6.52 – 0.74 -1.61 0.115
Offline FC 1.02 -1.64 – 3.68 0.78 0.442
Incentives -0.14 -0.26 – -0.02 -2.31 0.026
RSFC change -1.51 -4.54 – 1.52 -1.00 0.322
Incentives * RSFC change 3.60 -0.78 – 7.99 1.66 0.105
Observations 50
R2 / R2 adjusted 0.171 / 0.077

Smoothing kernel FWHM = 0: Linear model predicting Corrected memory

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 14.53 12.33 – 16.72 13.35 <0.001 14.26 12.25 – 16.27 14.74 <0.001 13.45 10.95 – 15.95 11.18 <0.001
Online FC -41.29 -133.55 – 50.97 -0.90 0.372 -10.07 -122.86 – 102.72 -0.19 0.854 -85.06 -256.13 – 86.00 -1.03 0.313
Offline FC 26.63 -40.95 – 94.20 0.79 0.431 10.43 -76.22 – 97.07 0.25 0.805 38.92 -76.39 – 154.22 0.70 0.490
Incentives -0.94 -4.05 – 2.17 -0.61 0.546
RSFC change -52.33 -129.40 – 24.75 -1.37 0.178 -47.22 -120.93 – 26.48 -1.33 0.197 78.72 -20.03 – 177.47 1.66 0.112
Incentives * RSFC change 124.73 13.28 – 236.18 2.26 0.029
Observations 50 25 25
R2 / R2 adjusted 0.117 / 0.016 0.080 / -0.052 0.151 / 0.030
  Whole sample
Predictors Estimates CI Statistic p
Intercept 14.53 12.33 – 16.72 13.35 <0.001
Online FC -41.29 -133.55 – 50.97 -0.90 0.372
Offline FC 26.63 -40.95 – 94.20 0.79 0.431
Incentives -0.94 -4.05 – 2.17 -0.61 0.546
RSFC change -52.33 -129.40 – 24.75 -1.37 0.178
Incentives * RSFC change 124.73 13.28 – 236.18 2.26 0.029
Observations 50
R2 / R2 adjusted 0.117 / 0.016

Smoothing kernel FWHM = 0: Linear model predicting Within-person correlation

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.07 -0.01 – 0.14 1.79 0.080 0.06 -0.03 – 0.14 1.36 0.187 0.01 -0.06 – 0.08 0.35 0.733
Online FC -2.25 -5.48 – 0.98 -1.41 0.167 -0.60 -5.42 – 4.22 -0.26 0.798 -3.74 -8.41 – 0.93 -1.67 0.111
Offline FC 0.33 -2.03 – 2.70 0.28 0.777 -1.45 -5.16 – 2.25 -0.82 0.423 2.26 -0.88 – 5.41 1.50 0.150
Incentives -0.06 -0.17 – 0.05 -1.09 0.282
RSFC change -2.02 -4.71 – 0.68 -1.51 0.139 -1.49 -4.64 – 1.66 -0.99 0.335 0.91 -1.78 – 3.61 0.70 0.489
Incentives * RSFC change 3.02 -0.88 – 6.92 1.56 0.126
Observations 50 25 25
R2 / R2 adjusted 0.104 / 0.003 0.129 / 0.005 0.168 / 0.049
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.07 -0.01 – 0.14 1.79 0.080
Online FC -2.25 -5.48 – 0.98 -1.41 0.167
Offline FC 0.33 -2.03 – 2.70 0.28 0.777
Incentives -0.06 -0.17 – 0.05 -1.09 0.282
RSFC change -2.02 -4.71 – 0.68 -1.51 0.139
Incentives * RSFC change 3.02 -0.88 – 6.92 1.56 0.126
Observations 50
R2 / R2 adjusted 0.104 / 0.003

Smoothing kernel FWHM = 4: Linear model predicting Total # items encoded

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 15.75 13.61 – 17.88 14.88 <0.001 15.52 13.56 – 17.48 16.50 <0.001 14.76 12.35 – 17.17 12.72 <0.001
Online FC -20.10 -103.75 – 63.55 -0.48 0.631 15.36 -85.45 – 116.16 0.32 0.755 -74.39 -229.93 – 81.14 -0.99 0.331
Offline FC 16.59 -43.56 – 76.75 0.56 0.581 0.15 -78.35 – 78.65 0.00 0.997 30.12 -68.41 – 128.64 0.64 0.532
Incentives -0.77 -3.80 – 2.26 -0.51 0.611
RSFC change -49.44 -118.33 – 19.46 -1.45 0.155 -45.68 -111.28 – 19.92 -1.45 0.162 84.79 -2.77 – 172.35 2.01 0.057
Incentives * RSFC change 127.24 28.24 – 226.23 2.59 0.013
Observations 50 25 25
R2 / R2 adjusted 0.143 / 0.046 0.104 / -0.024 0.194 / 0.078
  Whole sample
Predictors Estimates CI Statistic p
Intercept 15.75 13.61 – 17.88 14.88 <0.001
Online FC -20.10 -103.75 – 63.55 -0.48 0.631
Offline FC 16.59 -43.56 – 76.75 0.56 0.581
Incentives -0.77 -3.80 – 2.26 -0.51 0.611
RSFC change -49.44 -118.33 – 19.46 -1.45 0.155
Incentives * RSFC change 127.24 28.24 – 226.23 2.59 0.013
Observations 50
R2 / R2 adjusted 0.143 / 0.046

Smoothing kernel FWHM = 4: Linear model predicting Curiosity-driven memory benefit

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.10 0.01 – 0.18 2.27 0.028 0.08 -0.00 – 0.17 2.02 0.057 -0.05 -0.13 – 0.04 -1.10 0.284
Online FC -3.01 -6.34 – 0.32 -1.82 0.075 -1.70 -6.13 – 2.73 -0.80 0.435 -4.00 -9.50 – 1.50 -1.51 0.145
Offline FC 0.82 -1.58 – 3.21 0.69 0.494 -1.15 -4.60 – 2.30 -0.69 0.496 2.86 -0.62 – 6.35 1.71 0.102
Incentives -0.14 -0.26 – -0.02 -2.35 0.023
RSFC change -1.32 -4.07 – 1.42 -0.97 0.336 -0.81 -3.69 – 2.08 -0.58 0.567 1.99 -1.11 – 5.08 1.34 0.196
Incentives * RSFC change 3.63 -0.31 – 7.58 1.86 0.070
Observations 50 25 25
R2 / R2 adjusted 0.197 / 0.106 0.132 / 0.008 0.218 / 0.107
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.10 0.01 – 0.18 2.27 0.028
Online FC -3.01 -6.34 – 0.32 -1.82 0.075
Offline FC 0.82 -1.58 – 3.21 0.69 0.494
Incentives -0.14 -0.26 – -0.02 -2.35 0.023
RSFC change -1.32 -4.07 – 1.42 -0.97 0.336
Incentives * RSFC change 3.63 -0.31 – 7.58 1.86 0.070
Observations 50
R2 / R2 adjusted 0.197 / 0.106

Smoothing kernel FWHM = 4: Linear model predicting Corrected memory

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 14.47 12.27 – 16.68 13.22 <0.001 14.26 12.24 – 16.28 14.71 <0.001 13.45 10.93 – 15.97 11.10 <0.001
Online FC -22.54 -109.06 – 63.99 -0.52 0.602 6.62 -97.32 – 110.55 0.13 0.896 -68.65 -231.03 – 93.72 -0.88 0.389
Offline FC 21.76 -40.46 – 83.98 0.70 0.485 10.18 -70.75 – 91.11 0.26 0.796 30.67 -72.18 – 133.53 0.62 0.542
Incentives -0.86 -4.00 – 2.27 -0.56 0.581
RSFC change -43.83 -115.09 – 27.43 -1.24 0.222 -41.28 -108.91 – 26.36 -1.27 0.218 70.28 -21.13 – 161.70 1.60 0.125
Incentives * RSFC change 107.53 5.14 – 209.93 2.12 0.040
Observations 50 25 25
R2 / R2 adjusted 0.106 / 0.004 0.075 / -0.057 0.139 / 0.016
  Whole sample
Predictors Estimates CI Statistic p
Intercept 14.47 12.27 – 16.68 13.22 <0.001
Online FC -22.54 -109.06 – 63.99 -0.52 0.602
Offline FC 21.76 -40.46 – 83.98 0.70 0.485
Incentives -0.86 -4.00 – 2.27 -0.56 0.581
RSFC change -43.83 -115.09 – 27.43 -1.24 0.222
Incentives * RSFC change 107.53 5.14 – 209.93 2.12 0.040
Observations 50
R2 / R2 adjusted 0.106 / 0.004

Smoothing kernel FWHM = 4: Linear model predicting Within-person correlation

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.07 -0.01 – 0.14 1.83 0.074 0.06 -0.03 – 0.14 1.38 0.182 0.01 -0.06 – 0.08 0.35 0.733
Online FC -2.35 -5.32 – 0.63 -1.59 0.119 -1.07 -5.44 – 3.31 -0.51 0.617 -3.57 -7.98 – 0.85 -1.68 0.108
Offline FC 0.36 -1.78 – 2.50 0.34 0.735 -1.21 -4.61 – 2.20 -0.74 0.469 1.96 -0.84 – 4.75 1.46 0.160
Incentives -0.06 -0.17 – 0.05 -1.12 0.268
RSFC change -1.90 -4.35 – 0.55 -1.56 0.126 -1.49 -4.34 – 1.36 -1.09 0.288 0.90 -1.58 – 3.38 0.75 0.460
Incentives * RSFC change 2.96 -0.56 – 6.48 1.69 0.097
Observations 50 25 25
R2 / R2 adjusted 0.124 / 0.024 0.151 / 0.030 0.164 / 0.045
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.07 -0.01 – 0.14 1.83 0.074
Online FC -2.35 -5.32 – 0.63 -1.59 0.119
Offline FC 0.36 -1.78 – 2.50 0.34 0.735
Incentives -0.06 -0.17 – 0.05 -1.12 0.268
RSFC change -1.90 -4.35 – 0.55 -1.56 0.126
Incentives * RSFC change 2.96 -0.56 – 6.48 1.69 0.097
Observations 50
R2 / R2 adjusted 0.124 / 0.024

Smoothing kernel FWHM = 6: Linear model predicting Total # items encoded

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 15.70 13.66 – 17.75 15.47 <0.001 15.52 13.60 – 17.44 16.78 <0.001 14.76 12.45 – 17.07 13.30 <0.001
Online FC -1.80 -53.75 – 50.15 -0.07 0.945 13.19 -47.78 – 74.16 0.45 0.657 -32.15 -138.54 – 74.24 -0.63 0.536
Offline FC 10.12 -27.38 – 47.62 0.54 0.589 0.79 -49.54 – 51.12 0.03 0.974 21.49 -39.35 – 82.33 0.73 0.471
Incentives -0.65 -3.55 – 2.25 -0.45 0.652
RSFC change -35.06 -78.33 – 8.21 -1.63 0.110 -35.03 -76.60 – 6.54 -1.75 0.094 64.17 11.15 – 117.20 2.52 0.020
Incentives * RSFC change 94.35 33.95 – 154.74 3.15 0.003
Observations 50 25 25
R2 / R2 adjusted 0.207 / 0.117 0.133 / 0.010 0.263 / 0.158
  Whole sample
Predictors Estimates CI Statistic p
Intercept 15.70 13.66 – 17.75 15.47 <0.001
Online FC -1.80 -53.75 – 50.15 -0.07 0.945
Offline FC 10.12 -27.38 – 47.62 0.54 0.589
Incentives -0.65 -3.55 – 2.25 -0.45 0.652
RSFC change -35.06 -78.33 – 8.21 -1.63 0.110
Incentives * RSFC change 94.35 33.95 – 154.74 3.15 0.003
Observations 50
R2 / R2 adjusted 0.207 / 0.117

Smoothing kernel FWHM = 6: Linear model predicting Curiosity-driven memory benefit

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.09 0.00 – 0.18 2.10 0.042 0.08 -0.00 – 0.17 2.00 0.058 -0.05 -0.14 – 0.05 -1.03 0.313
Online FC -1.36 -3.54 – 0.83 -1.25 0.218 -0.83 -3.58 – 1.91 -0.63 0.533 -1.72 -5.90 – 2.47 -0.85 0.403
Offline FC 0.14 -1.44 – 1.72 0.18 0.861 -0.77 -3.03 – 1.50 -0.70 0.490 1.00 -1.39 – 3.40 0.87 0.393
Incentives -0.13 -0.25 – -0.01 -2.18 0.035
RSFC change -0.28 -2.10 – 1.54 -0.31 0.758 -0.13 -2.00 – 1.74 -0.14 0.889 1.41 -0.67 – 3.50 1.41 0.173
Incentives * RSFC change 1.80 -0.74 – 4.35 1.43 0.160
Observations 50 25 25
R2 / R2 adjusted 0.168 / 0.074 0.122 / -0.003 0.115 / -0.012
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.09 0.00 – 0.18 2.10 0.042
Online FC -1.36 -3.54 – 0.83 -1.25 0.218
Offline FC 0.14 -1.44 – 1.72 0.18 0.861
Incentives -0.13 -0.25 – -0.01 -2.18 0.035
RSFC change -0.28 -2.10 – 1.54 -0.31 0.758
Incentives * RSFC change 1.80 -0.74 – 4.35 1.43 0.160
Observations 50
R2 / R2 adjusted 0.168 / 0.074

Smoothing kernel FWHM = 6: Linear model predicting Corrected memory

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 14.43 12.30 – 16.56 13.68 <0.001 14.26 12.28 – 16.24 14.98 <0.001 13.45 11.02 – 15.88 11.50 <0.001
Online FC -2.75 -56.75 – 51.26 -0.10 0.919 7.27 -55.48 – 70.01 0.24 0.812 -24.68 -136.79 – 87.44 -0.46 0.652
Offline FC 12.14 -26.84 – 51.12 0.63 0.533 7.28 -44.51 – 59.08 0.29 0.773 18.62 -45.49 – 82.74 0.60 0.552
Incentives -0.73 -3.75 – 2.28 -0.49 0.627
RSFC change -32.05 -77.03 – 12.93 -1.44 0.158 -32.39 -75.18 – 10.39 -1.57 0.130 55.65 -0.23 – 111.53 2.07 0.051
Incentives * RSFC change 83.78 21.00 – 146.56 2.69 0.010
Observations 50 25 25
R2 / R2 adjusted 0.164 / 0.069 0.109 / -0.018 0.198 / 0.083
  Whole sample
Predictors Estimates CI Statistic p
Intercept 14.43 12.30 – 16.56 13.68 <0.001
Online FC -2.75 -56.75 – 51.26 -0.10 0.919
Offline FC 12.14 -26.84 – 51.12 0.63 0.533
Incentives -0.73 -3.75 – 2.28 -0.49 0.627
RSFC change -32.05 -77.03 – 12.93 -1.44 0.158
Incentives * RSFC change 83.78 21.00 – 146.56 2.69 0.010
Observations 50
R2 / R2 adjusted 0.164 / 0.069

Smoothing kernel FWHM = 6: Linear model predicting Within-person correlation

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.06 -0.01 – 0.14 1.69 0.099 0.06 -0.03 – 0.14 1.36 0.187 0.01 -0.06 – 0.08 0.33 0.746
Online FC -1.18 -3.14 – 0.78 -1.21 0.231 -0.60 -3.32 – 2.12 -0.46 0.653 -1.76 -5.08 – 1.56 -1.10 0.282
Offline FC 0.11 -1.30 – 1.53 0.16 0.871 -0.74 -2.99 – 1.50 -0.69 0.499 0.96 -0.94 – 2.85 1.05 0.306
Incentives -0.05 -0.16 – 0.06 -1.00 0.324
RSFC change -0.80 -2.43 – 0.83 -0.99 0.329 -0.67 -2.52 – 1.19 -0.75 0.464 0.59 -1.06 – 2.25 0.75 0.463
Incentives * RSFC change 1.44 -0.83 – 3.72 1.28 0.208
Observations 50 25 25
R2 / R2 adjusted 0.091 / -0.012 0.131 / 0.007 0.077 / -0.055
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.06 -0.01 – 0.14 1.69 0.099
Online FC -1.18 -3.14 – 0.78 -1.21 0.231
Offline FC 0.11 -1.30 – 1.53 0.16 0.871
Incentives -0.05 -0.16 – 0.06 -1.00 0.324
RSFC change -0.80 -2.43 – 0.83 -0.99 0.329
Incentives * RSFC change 1.44 -0.83 – 3.72 1.28 0.208
Observations 50
R2 / R2 adjusted 0.091 / -0.012

Smoothing kernel FWHM = 8: Linear model predicting Total # items encoded

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 15.68 13.67 – 17.70 15.71 <0.001 15.52 13.59 – 17.45 16.70 <0.001 14.76 12.52 – 17.00 13.71 <0.001
Online FC -1.07 -41.32 – 39.18 -0.05 0.958 10.24 -38.01 – 58.50 0.44 0.663 -26.23 -109.02 – 56.56 -0.66 0.517
Offline FC 6.85 -22.61 – 36.31 0.47 0.642 0.19 -40.34 – 40.72 0.01 0.992 15.86 -31.37 – 63.09 0.70 0.493
Incentives -0.62 -3.47 – 2.24 -0.44 0.665
RSFC change -25.68 -58.79 – 7.43 -1.56 0.125 -26.92 -59.42 – 5.58 -1.72 0.100 54.25 14.49 – 94.02 2.84 0.010
Incentives * RSFC change 75.42 30.06 – 120.78 3.35 0.002
Observations 50 25 25
R2 / R2 adjusted 0.231 / 0.143 0.125 / 0.000 0.306 / 0.207
  Whole sample
Predictors Estimates CI Statistic p
Intercept 15.68 13.67 – 17.70 15.71 <0.001
Online FC -1.07 -41.32 – 39.18 -0.05 0.958
Offline FC 6.85 -22.61 – 36.31 0.47 0.642
Incentives -0.62 -3.47 – 2.24 -0.44 0.665
RSFC change -25.68 -58.79 – 7.43 -1.56 0.125
Incentives * RSFC change 75.42 30.06 – 120.78 3.35 0.002
Observations 50
R2 / R2 adjusted 0.231 / 0.143

Smoothing kernel FWHM = 8: Linear model predicting Curiosity-driven memory benefit

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.09 -0.00 – 0.17 2.01 0.050 0.08 -0.00 – 0.17 1.99 0.060 -0.05 -0.14 – 0.05 -1.01 0.322
Online FC -0.58 -2.31 – 1.16 -0.67 0.505 -0.52 -2.70 – 1.66 -0.50 0.625 -0.13 -3.55 – 3.30 -0.08 0.939
Offline FC -0.16 -1.43 – 1.11 -0.25 0.806 -0.56 -2.39 – 1.27 -0.64 0.530 0.14 -1.82 – 2.09 0.15 0.885
Incentives -0.13 -0.25 – -0.00 -2.09 0.043
RSFC change -0.09 -1.52 – 1.34 -0.13 0.901 0.02 -1.45 – 1.48 0.02 0.983 0.95 -0.69 – 2.60 1.21 0.241
Incentives * RSFC change 1.23 -0.72 – 3.19 1.27 0.211
Observations 50 25 25
R2 / R2 adjusted 0.155 / 0.058 0.108 / -0.019 0.079 / -0.053
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.09 -0.00 – 0.17 2.01 0.050
Online FC -0.58 -2.31 – 1.16 -0.67 0.505
Offline FC -0.16 -1.43 – 1.11 -0.25 0.806
Incentives -0.13 -0.25 – -0.00 -2.09 0.043
RSFC change -0.09 -1.52 – 1.34 -0.13 0.901
Incentives * RSFC change 1.23 -0.72 – 3.19 1.27 0.211
Observations 50
R2 / R2 adjusted 0.155 / 0.058

Smoothing kernel FWHM = 8: Linear model predicting Corrected memory

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 14.42 12.32 – 16.51 13.88 <0.001 14.26 12.28 – 16.24 14.97 <0.001 13.45 11.08 – 15.82 11.81 <0.001
Online FC -1.87 -43.74 – 40.00 -0.09 0.929 5.93 -43.54 – 55.41 0.25 0.805 -20.88 -108.47 – 66.71 -0.50 0.625
Offline FC 8.29 -22.36 – 38.94 0.55 0.588 4.76 -36.80 – 46.31 0.24 0.814 13.80 -36.17 – 63.77 0.57 0.572
Incentives -0.70 -3.67 – 2.27 -0.47 0.637
RSFC change -24.05 -58.49 – 10.39 -1.41 0.166 -25.22 -58.54 – 8.11 -1.57 0.130 48.01 5.94 – 90.08 2.37 0.027
Incentives * RSFC change 68.34 21.16 – 115.53 2.92 0.006
Observations 50 25 25
R2 / R2 adjusted 0.188 / 0.095 0.107 / -0.020 0.239 / 0.130
  Whole sample
Predictors Estimates CI Statistic p
Intercept 14.42 12.32 – 16.51 13.88 <0.001
Online FC -1.87 -43.74 – 40.00 -0.09 0.929
Offline FC 8.29 -22.36 – 38.94 0.55 0.588
Incentives -0.70 -3.67 – 2.27 -0.47 0.637
RSFC change -24.05 -58.49 – 10.39 -1.41 0.166
Incentives * RSFC change 68.34 21.16 – 115.53 2.92 0.006
Observations 50
R2 / R2 adjusted 0.188 / 0.095

Smoothing kernel FWHM = 8: Linear model predicting Within-person correlation

  Whole sample Control group Incentives group
Predictors Estimates CI Statistic p Estimates CI Statistic p Estimates CI Statistic p
Intercept 0.06 -0.02 – 0.14 1.60 0.117 0.06 -0.03 – 0.14 1.36 0.190 0.01 -0.06 – 0.09 0.32 0.753
Online FC -0.50 -2.06 – 1.05 -0.65 0.519 -0.34 -2.50 – 1.82 -0.33 0.746 -0.29 -3.03 – 2.46 -0.22 0.830
Offline FC -0.13 -1.27 – 1.01 -0.23 0.818 -0.59 -2.40 – 1.22 -0.68 0.505 0.24 -1.33 – 1.81 0.32 0.753
Incentives -0.05 -0.16 – 0.06 -0.91 0.367
RSFC change -0.48 -1.76 – 0.80 -0.76 0.454 -0.39 -1.84 – 1.06 -0.56 0.584 0.32 -1.00 – 1.64 0.50 0.620
Incentives * RSFC change 0.94 -0.81 – 2.70 1.08 0.285
Observations 50 25 25
R2 / R2 adjusted 0.071 / -0.034 0.120 / -0.006 0.018 / -0.123
  Whole sample
Predictors Estimates CI Statistic p
Intercept 0.06 -0.02 – 0.14 1.60 0.117
Online FC -0.50 -2.06 – 1.05 -0.65 0.519
Offline FC -0.13 -1.27 – 1.01 -0.23 0.818
Incentives -0.05 -0.16 – 0.06 -0.91 0.367
RSFC change -0.48 -1.76 – 0.80 -0.76 0.454
Incentives * RSFC change 0.94 -0.81 – 2.70 1.08 0.285
Observations 50
R2 / R2 adjusted 0.071 / -0.034